The pre-market tape told a divided story on July 10. Meta was up more than 4 percent, leading the technology sector, while the chip stocks that have powered the AI rally were broadly lower, and the day’s focal point, the first trading session of SK Hynix’s American listing, hung over both halves of the market.
36Kr’s pre-market data showed the split in detail. Meta rose more than 4 percent on the strength of its AI narrative, Microsoft gained nearly 1 percent, and Amazon added 0.45 percent. On the other side, Tesla fell 0.38 percent, Apple dropped 0.53 percent, Nvidia lost 0.5 percent and Alphabet declined 0.38 percent. The divergence was not random; it tracked the day’s two competing stories about the economics of AI.
Meta’s jump had two engines. The first was Muse Spark 1.1, the model the company launched on July 10 with aggressive pricing for developers, which traders read as the beginning of a real monetization effort for Meta’s AI work. The second was the MTIA chip program, Meta’s in-house silicon effort, which has become a bull story for the stock: investors see a company that is both selling AI to developers and reducing its dependence on Nvidia’s prices.
The chip weakness was the mirror image. Nvidia’s decline was attributed by traders to the same MTIA narrative, with Barron’s having published an analysis the previous day pointing to Meta’s custom chips as a pressure on Nvidia’s data center business. The logic cuts both ways in real time: the story that lifts Meta’s stock is the story that weighs on Nvidia’s, and the pre-market tape captured both sides of that trade in a single morning.
The debut itself carried the day’s largest question. A $26.5 billion listing that was oversubscribed, with institutional orders above $200 million and single American funds committing more than $1 billion, had set expectations high, and the first session’s trading would show whether the demand in the book matched the demand in the market. The memory maker’s outcome was being watched as a signal for the entire AI supply chain: if investors pay up for HBM capacity, the valuation logic of the semiconductor complex holds; if the debut fades, the chip bears gain their opening.
The SK Hynix debut tied the two halves together. The memory maker’s $26.5 billion listing, the largest ever for a Korean company, began trading on Nasdaq the same morning, and its outcome was being read as a referendum on the AI supply chain: the chips, the memory and the capital that flows into both. A strong debut would support the entire sector’s valuation logic; a weak one would give the chip bears their opening.
The rotation, if it persists, would have a coherent logic. The AI trade has run for two years on the assumption that the infrastructure buildout, the chips, the memory, the data centers, would capture the value of the technology. The counter-narrative, gaining force this week, holds that the value is migrating to the platforms that own the models and the distribution, and that the hardware names have been priced for a growth rate their customers are now preparing to produce themselves. Meta, with its model launch and its custom silicon, sits on the winning side of both narratives, which is why its stock led the morning.
Microsoft’s modest gain reflected the same AI-compute complex from a different angle. The company’s Azure business is a major route for OpenAI models into the enterprise, and its nearly 1 percent advance was attributed to the continued strength of that relationship, reinforced the previous day by Sam Altman’s public commitment that Microsoft will remain one of OpenAI’s largest customers.
The laggards had no single catalyst. Tesla’s decline was attributed by traders to the sector rotation out of consumer-facing names, Apple’s to the absence of any fresh AI narrative and Alphabet’s to the ongoing cost concerns around its AI search products. None of the moves was dramatic, and analysts said the pattern reflected positioning ahead of the SK Hynix debut and the broader chip-sector repricing rather than company-specific news.
The session’s real question, market participants said, was whether the chip weakness would spread. The AI trade has been the market’s center of gravity for two years, and a rotation out of the hardware names into the platform names, with Meta the clearest beneficiary, would mark a shift in how investors express their AI conviction. The pre-market action offered an early version of that rotation, and traders were watching whether it survived the opening bell.
By the time the market opened, the shape of the day was set: Meta leading, chips lagging, and the memory stock debut as the backdrop. The split captured the industry’s central tension, between the companies that build the AI infrastructure and the companies that own the platforms running on top of it, and the market was pricing both sides at once. Whether the divergence becomes a trend will be decided in the sessions ahead, as the SK Hynix trading pattern and the chip sector’s reaction to the custom-silicon story take their course.
